Google DeepMind Is Losing Its Grip on Elite AI Talent, New Data Shows

For most of the past decade, Google DeepMind was the destination for elite AI researchers — especially in Europe, where the London-born lab hoovered up roughly half of all top-tier research and engineering hires. New data shared exclusively with Fortune shows that grip has now slipped dramatically, and the numbers behind the slide are starker than the headline departures suggest.

The analysis, from UK-based talent-intelligence firm Zeki Data, tracks 20,900 people in research and advanced-engineering roles across ten major AI companies, compiled from publicly available information as of August 2026. Its central finding: where OpenAI and Anthropic are making gains in the AI talent market, Google DeepMind — and to some extent Meta — are stumbling.

The numbers behind the slide

The headline statistic is regional. DeepMind’s share of research and advanced-engineering hires across Europe, the Middle East and Africa (EMEA) fell from 49% in 2022–23 to just 18.6% in 2025–26. Zeki describes this as the sharpest market-share drop it recorded for a major AI lab in any region of the world.

“They had the crown in Europe forever, and then it started to erode from a very high base,” Zeki Data founder Tom Hurd told Fortune. “The likes of Microsoft AI Superintelligence and Meta Superintelligence are eating into their market share, and then there’s OpenAI and Anthropic on the side.”

Globally, DeepMind is still bringing in more research and advanced-engineering staff than it loses — but the margin has thinned dramatically. The lab’s arrivals-to-departures ratio has fallen from roughly 12-to-1 in the second quarter of 2023 to about 2-to-1 in the third quarter of 2026. In other words, it now adds two people for every one who walks out the door, where it once added twelve. For comparison, Meta’s ratio in 2025 was 3-to-1, OpenAI’s was 5.7-to-1, and Anthropic’s was 22-to-1.

Anthropic has become the single leading destination for departing DeepMind researchers. Of those who left the lab in the past 12 months, 25% went to Anthropic, 21% to Meta, and 14% to OpenAI, according to Zeki’s data. Representatives for Google DeepMind did not respond to a request for comment on the findings.

Growth is flowing to newer labs

Sector-wide, research and engineering hiring has grown at a compound annual rate of 23% since 2022 — but that growth is flowing disproportionately to the newer frontier labs. OpenAI’s research and engineering headcount has grown at roughly 97% annually, and Anthropic’s at 152%, compared with 27% for Google DeepMind. OpenAI and Anthropic started from a much smaller base, but the divergence shows how quickly those organizations have expanded relative to the incumbent.

DeepMind is also facing new competition in regions it long dominated. Mistral AI and Anthropic have been the primary beneficiaries of its declining EMEA share. In Asia-Pacific — where DeepMind opened a new research lab in Singapore last November — domestic players including ByteDance, Sakana AI and Sarvam AI are building share in markets the incumbent labs historically underinvested in.

Where the losses hurt most

Zeki’s data suggests the attrition is not evenly spread. Comparing the expertise of everyone who left DeepMind with those who joined during the past 12 months, the firm found a net loss specifically in large language models and multimodal systems: 19.2% of leavers specialized in those areas, versus 15.6% of joiners. The company has also lost ground in computer vision.

The AlphaFold team has been hit particularly hard. Of the 29 named authors on the AlphaFold2 paper — the breakthrough system that predicts protein structures from genetic sequences — 13 have left DeepMind since publication. The Financial Times reported in July that most of the original paper’s authors had been reassigned over the preceding year, moving to Gemini-related projects, enzyme design, genomics, nuclear fusion, and Isomorphic Labs, Alphabet’s drug-discovery subsidiary.

There is one bright spot: DeepMind is gaining ground in robotics, embodied AI, and machine learning for science — fields where it is expanding capacity and, in some cases, competing for talent with Nvidia as much as with rival AI labs.

Why researchers are leaving

Interviews with three current and two former DeepMind staffers, conducted by Fortune, point to a converging set of pressures. Rival labs like Meta and Microsoft have been aggressively poaching with cash-heavy offers. Frustration inside Google over where it stands in the AI race has dented morale. And the pull of pre-IPO equity at OpenAI and Anthropic — packages that could mint early employees into millionaires if those companies go public — is increasingly hard to resist.

But the deeper issue may be identity. As Google has pushed to close the gap with OpenAI and Anthropic, DeepMind has become more tightly organized around improving and commercializing Gemini — a shift some researchers see as displacing the open-ended, long-horizon science that originally made the lab such a desirable destination for academics.

The lab’s tightening publication rules have added friction. The Financial Times first reported in April 2025 that DeepMind had introduced a tougher internal review process and a six-month embargo for some strategically sensitive generative-AI papers, aiming to keep competitors from benefiting from its research. For a lab that built its reputation on public breakthroughs like AlphaGo and AlphaFold, the shift grated on researchers who had signed up for relatively unconstrained, blue-sky work.

“Most old timers who joined DeepMind before the ChatGPT moment, joined to be part of an AI research lab,” one former DeepMind engineer told Fortune. “Anyone who joined before 2023 thought they were joining an AI research lab, and suddenly they were asked to build products for Google.”

A summer of high-profile exits

The data lands on top of an unusually bruising stretch for the lab. This month alone, Google DeepMind lost Jeff Dean, the company’s longtime chief scientist and a 27-year veteran, alongside senior fellow Sanjay Ghemawat and researchers Oriol Vinyals and Quoc Le, who left to launch Discovery Loop — a startup reportedly in talks for a $1 billion funding round. On the same afternoon, DeepMind cofounder and CEO Demis Hassabis announced he would step back from day-to-day control to become the lab’s chairman and Alphabet’s chief scientist, handing operational control to CTO Koray Kavukcuoglu.

Earlier in the summer, the departures came even faster. In June, Google lost Gemini co-lead Noam Shazeer to OpenAI and John Jumper — who shared the 2024 Nobel Prize in Chemistry for AlphaFold with Hassabis — to Anthropic within a day of one another. Fellow AlphaFold researchers Jonas Adler and Alexander Pritzel followed Jumper to Anthropic shortly afterward. David Silver, the reinforcement-learning pioneer behind AlphaGo, AlphaZero and AlphaStar, left after nearly 13 years to launch Ineffable Intelligence, a London startup now valued at $5.1 billion after a $1.1 billion seed round.

“The more research-heavy people feel that they’d quite like to look at opportunities elsewhere, and then you start to see a much larger number than usual going to set up their own thing and bringing some of their friends with them,” Hurd said. “Over time, that’s added up to a large number of people.”

What it means

None of this means DeepMind is collapsing. The lab still commands Google’s compute, cash, reputation, and institutional reach — advantages almost no other AI lab can match. It remains a net importer of talent, and its gains in robotics and AI-for-science suggest it is deliberately reallocating toward areas where it sees the next decade of leverage.

But the trend line matters in a market where a relatively small pool of researchers and engineers has the experience to train and improve frontier models. Their choices determine how quickly a lab’s models improve, whether research breakthroughs become products, and how credibly it can attract the next wave of talent. In that contest — where the best researchers can choose almost anywhere and work on almost anything — DeepMind’s old advantages are proving harder to preserve than anyone at Google expected.